Assessment of road-cut slope stability using empirical, numerical, and machine learning methodologies
Abstract Landslides represent one of the most widespread natural hazards affecting mountainous regions globally. Their occurrence is primarily attributed to natural factors such as high-intensity precipitation, steeper topography, and seismic activity. These factors are further exacerbated by anthro...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-06-01
|
| Series: | Discover Civil Engineering |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44290-025-00267-z |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Landslides represent one of the most widespread natural hazards affecting mountainous regions globally. Their occurrence is primarily attributed to natural factors such as high-intensity precipitation, steeper topography, and seismic activity. These factors are further exacerbated by anthropogenic activities, including unregulated infrastructural developments, mining operations, and the expansion of transportation infrastructure such as roads and railways. Understanding landslide susceptibility is essential for sustainable socio-economic development in mountainous regions. This study aims to conduct a systematic comparison of empirical, numerical, and machine learning approaches for slope stability assessment, with the objective of identifying their respective strengths, limitations, and real-time applications. The critical parameters involved in slope stability analyses have been collected through extensive field visits. The slope stability analysis encompasses empirical approaches, including rock mass classification methods and kinematic analysis; numerical procedures such as Limit Equilibrium Analysis (LEA) and Finite Element Analysis (FEA); and machine learning techniques, specifically Random Forest (RF), Gradient Boosting Machine (GBM), and XGBoost. These methods are applied to seven road-cut slopes situated along a proposed tunnel path in a geologically complex mountainous region. The objective is to evaluate the effectiveness of each approach and to assess the potential of advanced machine learning models as viable validation tools for conventional slope stability assessment outcomes. The study reveals that the surveyed locations exhibit varying stability grades as determined by Rock Mass Classification systems and corresponding Factor of Safety (FOS) values. The outcomes of the machine learning algorithms are compared against those obtained from Finite Element Analysis (FEA), with observed discrepancies ranging from 2 to 10% across the different ML models. |
|---|---|
| ISSN: | 2948-1546 |